University of S˜aoPaulo College of Economics, Business and Accounting (FEA/USP) Dept. of Accountancy and Actuarial Science Marcelo dos Santos Guzella Investor Attention in the Brazilian Stock Market: Essays in Behavioral Finance Atenc¸ao˜ do Investidor no Mercado Brasileiro de Ac¸oes:˜ Ensaios em Financ¸as Comportamentais S˜aoPaulo 2020 Prof. DSc. Vahan Agopyan Dean of the University of S˜aoPaulo Prof. DSc. F´abioFrezatti Director of the College of Economics, Business and Accounting (FEA/USP) Prof. DSc. Valmor Slomski Chief Officer of the Dept. of Accountancy and Actuarial Science Prof. DSc. Lucas Ayres Barreira de Campos Barros Coordinator of the Postgraduate Program in Controllership and Accountancy Marcelo dos Santos Guzella Investor Attention in the Brazilian Stock Market: Essays in Behavioral Finance Dissertation presented to the Dept. of Accoun- tancy and Actuarial Science of the College of Eco- nomics, Business and Accounting (FEA/USP) of the University of S˜aoPaulo as partial requirement for obtaining the title of Doctor of Science. Supervisor: Prof. DSc. Francisco Henrique Figueiredo de Castro Junior Final version S˜aoPaulo 2020 I authorize the total or partial reproduction and disclosure of this material, by any con- ventional or electronic means, for the purposes of study and research, provided the source is cited. Catalog Card Prepared by the Dept. of Technical Processing of SBD/FEA/USP Guzella, Marcelo dos Santos. Investor Attention in the Brazilian Stock Market: Essays in Be- havioral Finance / Marcelo dos Santos Guzella. – S˜aoPaulo, 2020. 114 p. Dissertation (Doctoral Program) – University of S˜aoPaulo, 2020. Supervisor: Francisco Henrique Figueiredo de Castro Junior. 1. Investor Attention. 2. Behavioral Finance. 3. Internet Search Volume. 4. Stock Markets. 5. Volatility. I. University of S˜aoPaulo. College of Economics, Business and Accounting (FEA/USP). II. Title. This study was financed in part by the Coordination for the Improvement of Higher Education Personnel (CAPES) - Finance Code 001. To my family. iii iv Acknowledgements It is an impossible task to mention exhaustively all those who have somehow contributed to this meaningful accomplishment. First I would like to thank God for the opportunity, strength and ability to under- take this research and to conclude it reasonably. I also wish to express my deepest gratitude to my family for encouraging and supporting me during this work: my mother, Neusa, and particularly my father, Fernando, who while alive did not save efforts to encourage me to pursue my life aspirations; and my brothers, Thiago, Matheus and Rodrigo, whose academic achievements inspired me to carry out this challenging mission. I will not run the risk of mentioning other family members and friends who helped me on this journey, but here is my appreciation and consideration to all those ones whose assistance was a milestone in the completion of this project. I would like to pay my special regards to my Supervisor, Professor Henrique Castro, for the valuable ideas and assistance along the research process. Your great advice proved monumental towards the success of this study. I also thank Professor Veronica Santana for always being so open and patient while advising me on so many methodological decisions and issues. I also recognize the invaluable support of the Department Coordinator, Professor Lucas Barros, and all the other professors of the Department, particularly the ones with whom I had the privilege to have classes throughout the Doctoral Program. Moreover, I cannot forget the important support of the Department’s technical and support staff, as well as my doctoral colleagues. The experience at Columbia Business School was also of great value for this aca- demic achievement, with special thanks to my Visiting Supervisor Professor Kent Daniel and Professors Xavier Giroud and Stijn Van Nieuwerburgh. The professors of Coppead UFRJ and EM Lyon also deserve sincere thanks. Those are distinguished Business Schools where I studied during the Master’s Program and got academic and methodological bases that were fundamental for conducting this research. I wish to show my gratitude to the Professors that, together with my Supervisor, composed the Qualification Board, Lucas Barros, Ricardo Rochman and Andrea Minardi, as well as the the participants of the Brazilian Conference of Behavioral Finance and v vi the Brazilian Finance Meeting, who gave rich comments and suggestions to improve this research endeavor. Last but not least, I am indebted to Codemge (Development Company of Minas Gerais) for the support needed to concile professional activities with the Doctoral Program and this dissertation, so I express my gratitude to former and current directors (specially Mr. Ricardo Toledo and Mr. Marco Antonio Castello Branco) and colleagues. Abstract Guzella, M. S. (2020). Investor Attention in the Brazilian Stock Market: Essays in Be- havioral Finance (PhD Dissertation, University of S˜aoPaulo, S˜aoPaulo). We developed three essays regarding the impact of the investor attention on the Brazilian stock market. Attention is a cognitive resource of great relevance and has been increas- ingly studied in research related to behavioral finance. It has a crucial role in processes such as buying and selling assets, absorption of information and risk management. Firstly, we evidenced that attention transmits market efficiency for being a requirement for the discovery of released information, and this effect is more associated with professional at- tention. After that, we verified that attention, particularly the one of retail investors, is capable of inducing volatility to the market due to a price pressure by noise trading. Finally, we verified that the volatility of prices is less asymmetric when investors are more attentive to financial information. We measured attention through the volume of searches for financial information over the Internet, particularly the queries performed using Google and Bloomberg. These indicators have properties that allow several ap- proaches that were limited or impossible before they were available. All the results are robust to different methodologies and specifications. Among other innovations, this study is a pioneer in isolating the effect of retail and professional attention on the market effi- ciency and asymmetry. Our findings contribute to a better understanding of the influence of aggregate psychological aspects on stock prices and open promising venues for research ideas in many fields. Keywords: Investor Attention, Behavioral Finance, Internet Search Volume, Stock Mar- kets, Volatility. vii Contents Acronyms 7 1 Introduction 9 1.1 Attention Variables . 13 1.1.1 Retail Investor Attention: Google Search Volume (GSVt) . 13 1.1.2 Professional Investor Attention: Bloomberg Search Volume (BSVt) 14 1.1.3 Isolated Non-Professional Attention: Google Search Volume Resid- uals (rGSVt) .............................. 15 References . 16 2 Information Discovery: Effects of Investor Attention on the Market Efficiency 19 2.1 Literature Review . 21 2.1.1 Information Discovery Hypothesis . 21 2.1.2 Competition for Attention . 26 2.1.3 Attention and Market Efficieny . 28 2.2 Methodology . 30 2.2.1 Data Description . 31 2.2.2 Modeling . 32 2.3 Empirical Results . 34 2.3.1 Descriptive Analysis . 34 2.3.2 Autoregressive Modeling . 37 2.3.3 Effects of Filtered Non-Professional Attention . 40 2.4 Final Considerations of the Chapter . 41 References . 43 3 Price Pressure Induction: Effects of the Investor Attention on the Mar- ket Volatility 47 3.1 Literature Review . 48 3.1.1 Behavioral Biases in Financial Markets . 48 3.1.2 Volatility Induction by Noise Traders . 49 3.1.3 Effects of the Attention Level on Financial Markets . 50 3.1.4 Internet Search Queries as an Indicator of Investor Attention . 52 3.1.5 Empirical Evidence about Investor Attention in Brazil . 56 3.1.6 Additional Considerations Regarding Attention and Financial Mar- kets ................................... 57 3.2 Methodology . 58 3.2.1 Data Description . 58 3.2.2 Modeling . 60 1 2 Contents 3.3 Empirical Results . 61 3.3.1 Descriptive Analysis . 61 3.3.2 Generalized AutoRegressive Conditional Heteroskedasticity (GARCH) Modeling . 62 3.3.3 Vector Autoregressive (VAR) Modeling . 65 3.4 Final Considerations of the Chapter . 68 References . 70 4 Ostrich Behavior: Effects of Investor Attention on the Volatility Asym- metry 73 4.1 Literature Review . 76 4.1.1 Volatility Asymmetry . 77 4.1.2 Attention Reaction to the Asset Prices . 81 4.2 Methodology . 87 4.2.1 Data Description . 87 4.2.2 Variance Asymmetry Modeling . 89 4.2.3 Modeling the Relationship between Asymmetry and Attention . 91 4.3 Empirical Results . 91 4.3.1 Variance Asymmetry Modeling Outcomes . 92 4.3.2 Outcomes of the Relationship between Asymmetry and Attention . 94 4.3.3 Isolating the Effects of Professional and Retail Attention on Asym- metry . 97 4.4 Final Considerations of the Chapter . 98 References . 100 5 Concluding Remarks 105 References 107 List of Tables 2.1 Sample of stocks . 33 2.2 Descriptive statistics of the time series . 35 2.3 Autoregressive estimation for AmBev (largest Brazilian beverage company) 38 2.4 Autoregressive estimation for CSN (large Brazilian steel company) . 39 2.5 Summary of the results for the stocks . 40 2.6 Results using rGSV (non-professional attention uncorrelated with profes- sional one) . 42 3.1 Correlation between “Bovespa Index (Ibovespa)”-related keywords . 62 3.2 Descriptive statistics of the time series . 62 3.3 GARCH modeling of the Ibovespa log-returns . 64 3.4 Correlation matrix of the series . 65 3.5 Vector autoregressive modeling . 67 4.1 Descriptive statistics . 89 4.2 Asymmetry Modeling of the Ibovespa . 93 4.3 Parameters estimation of the longitudinal determinants of Ibovespa daily volatility asymmetry . 95 4.4 Average asymmetry in each quartile of returns and attention . 96 4.5 Estimates of the effect of professional and retail attention on the asymmetry of the Ibovespa daily volatility . 99 3 4 List of Tables List of Figures 1.1 Search engines market share in Brazil .